U.S. patent application number 10/395264 was filed with the patent office on 2003-10-02 for method and system for target detection using an infra-red sensor.
This patent application is currently assigned to LOCKHEED MARTIN CORPORATION. Invention is credited to Chen, Hai-Wen, Frey, Steven R. JR., Olson, Teresa L..
Application Number | 20030183765 10/395264 |
Document ID | / |
Family ID | 28457152 |
Filed Date | 2003-10-02 |
United States Patent
Application |
20030183765 |
Kind Code |
A1 |
Chen, Hai-Wen ; et
al. |
October 2, 2003 |
Method and system for target detection using an infra-red
sensor
Abstract
A target detection and tracking system provides dynamic changing
of the integration time (IT) for the system IR sensor within a
discrete set of values to maintain a high sensor sensitivity. The
system changes the integration time to the same or a different
sensor integration time within the discrete set based on the image
data output from the sensor satisfying pre-determined system
parameter thresholds. The system includes an IT-related saturation
prediction function allowing the system to avoid unnecessary system
saturation when determining whether an IT change should be made.
The tracking portion of the system provides tracking feedback
allowing target objects with a low sensor signature to be detected
without being obscured by nearby uninterested objects that produce
system saturation.
Inventors: |
Chen, Hai-Wen; (Orlando,
FL) ; Frey, Steven R. JR.; (Orlando, FL) ;
Olson, Teresa L.; (Winter Garden, FL) |
Correspondence
Address: |
BIRCH STEWART KOLASCH & BIRCH
PO BOX 747
FALLS CHURCH
VA
22040-0747
US
|
Assignee: |
LOCKHEED MARTIN CORPORATION
Orlando
FL
|
Family ID: |
28457152 |
Appl. No.: |
10/395264 |
Filed: |
March 25, 2003 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
60367282 |
Mar 26, 2002 |
|
|
|
Current U.S.
Class: |
250/330 |
Current CPC
Class: |
G06K 9/6289 20130101;
G06K 9/6293 20130101; G06T 5/009 20130101; G06T 5/40 20130101; G06T
2207/10048 20130101; G06V 10/255 20220101; G06K 9/6288
20130101 |
Class at
Publication: |
250/330 |
International
Class: |
G06T 001/00 |
Claims
What is claimed is:
1. A method for dynamically changing a target detection system
parameter, comprising: receiving image data, output from a sensor
in a receiving system and generated using one of a discrete set of
sensor integration times wherein each integration time being
associated with a corresponding temperature range; determining
whether said image data satisfies at least one predetermined
threshold; and selecting the same one or a different sensor
integration time for a succeeding frame of image data to produce a
predetermined sensor sensitivity based on said image data
satisfying said at least one predetermined threshold.
2. The method of claim 1, wherein said selecting includes selecting
the same one or a different sensor integration time based on said
image data not producing saturation in the system.
3. The method of claim 2, wherein said selecting includes selecting
the same one sensor integration time based on predicting saturation
being produced in the receiving system from the succeeding frame of
said image data.
4. The method of claim 2, wherein said selecting includes selecting
a different sensor integration time based on predicting saturation
not being produced in the receiving system from a succeeding frame
of said image data.
5. The method of claim 1, wherein said selecting includes selecting
the same one sensor integration time based on said image data
producing saturation and target tracking in the system.
6. The method of claim 1, wherein said selecting includes selecting
the same one sensor integration time based on said image data
including non-target data producing saturation and no target
tracking in the system.
7. The method of claim 1, wherein said selecting includes selecting
a different sensor integration time based on said image data
including non-target data producing saturation and no target
tracking in the system.
8. The method of claim 1, wherein said receiving includes receiving
image data, output from a sensor in a receiving system and
generated using one of a discrete set of less than five sensor
integration times wherein each integration time being associated
with a corresponding temperature range.
9. The method of claim 1, wherein said selecting includes selecting
the same one or a different sensor integration time to produce a
pre-determined sensor sensitivity and sensor target detection range
based on said image data satisfying said at least one
pre-determined threshold.
10. A system for dynamically changing a target detection system
parameter, comprising: a sensor for receiving image data in a
receiving system and outputting image data using one of a discrete
set of sensor integration times wherein each integration time being
associated with a corresponding temperature range; and a
controller, interconnected to said sensor, for determining whether
said image data satisfies at least one predetermined threshold and
selecting the same one or a different sensor integration time for a
succeeding frame of image data to produce a predetermined sensor
sensitivity based on said image data satisfying said at least one
predetermined threshold.
11. The system of claim 10, wherein said controller to select the
same one or a different sensor integration time based on said image
data not producing saturation in the system.
12. The system of claim 11, wherein said controller to select the
same one sensor integration time based on predicting saturation
being produced in the receiving system from a succeeding frame of
said image data.
13. The system of claim 11, wherein said controller to select a
different sensor integration time based on predicting saturation
not being produced in the receiving system from a succeeding frame
of said image data.
14. The system of claim 10, wherein said controller to select the
same one sensor integration time based on said image data producing
saturation and target tracking in the system.
15. The system of claim 10, wherein said controller to select the
same one sensor integration time based on said image data including
non-target data producing saturation and no target tracking in the
system.
16. The system of claim 10, wherein said controller to select a
different sensor integration time based on said image data
including non-target data producing saturation and no target
tracking in the system.
17. The system of claim 10, wherein said sensor to output image
data using one of a discrete set of less than five sensor
integration times wherein each integration time being associated
with a corresponding temperature range.
18. The system of claim 10, wherein said controller to select the
same one or a different sensor integration time to produce a
predetermined sensor sensitivity and sensor target detection range
based on said image data satisfying said at least one predetermined
threshold.
19. A machine-readable medium having stored thereon a plurality of
executable instructions, the plurality of instructions comprising
instructions to: determine whether received image data, generated
and output from a sensor using one of a discrete set of sensor
integration times wherein each integration time being associated
with a corresponding temperature range, satisfies at least one
predetermined threshold; and select the same one or a different
sensor integration time for a succeeding frame of image data to
produce a predetermined sensor sensitivity based on said image data
satisfying said at least one predetermined threshold.
20. The medium of claim 19, wherein said plurality of executable
instructions include further instructions to select the same one or
a different sensor integration time based on said image data not
producing saturation in the system.
21. The system of claim 20, wherein said plurality of executable
instructions include further instructions to select the same one
sensor integration time based on predicting saturation being
produced in the receiving system from a succeeding frame of said
image data.
22. The system of claim 20, wherein said plurality of executable
instructions include further instructions to select a different
sensor integration time based on predicting saturation not being
produced in the receiving system from a succeeding frame of said
image data.
23. The system of claim 19, wherein said plurality of executable
instructions include further instructions to select the same one
sensor integration time based on said image data producing
saturation and target tracking in the system.
24. The system of claim 19, wherein said plurality of executable
instructions include further instructions to select the same one
sensor integration time based on said image data including
non-target data producing saturation and no target tracking in the
system.
25. The system of claim 19, wherein said plurality of executable
instructions include further instructions to output image data
using one of a discrete set of less than five sensor integration
times wherein each integration time being associated with a
corresponding temperature range.
26. The system of claim 19, wherein said plurality of executable
instructions include further instructions to select the same one or
a different sensor integration time to produce a predetermined
sensor sensitivity and sensor target detection range based on said
image data satisfying said at least one predetermined threshold.
Description
TECHNICAL FIELD
[0001] The present invention relates generally to image processing.
It particularly relates to an image processing target detection
system and method that dynamically changes at least one system
parameter to detect targets within various background temperature
ranges.
BACKGROUND OF THE INVENTION
[0002] IR (Infrared) sensors are widely used in current missile
targeting systems (e.g., self-targeting submissiles) to help detect
and track target objects in a cluttered background. However,
especially in poor weather conditions, the raw image (frame of
pixels) data output from the IR sensor may frequently contain a
significant portion of non-uniform/fixed pattern noise (FPN) in
addition to permanent and blinking dead pixels. These dead pixels
are either very bright or dark in intensity leading to non-linear
(e.g., saturation or starvation) conditions for the targeting
system resulting in low target detection reliability and lower
sensor sensitivity. Therefore, many missile targeting systems
include a NUC (non-uniform correction) system that attempts to
replace the dead pixels and/or reduce the FPN for more reliable
target detection (of target signature) and higher sensor
sensitivity.
[0003] FIG. 1 illustrates an exemplary target detection system 100
found in the prior art that attempts to eliminate and/or reduce the
noise and dead pixel problem. During operation, an IR sensor 102,
preferably including an FPA (focal plane array), receives the
radiant flux from the sensing area and outputs (generates) a raw
image data signal 105 (e.g., target signature), at an output
voltage (V.sub.p), to amplifier 106 using a capacitor circuit 104.
The amplifier 106 outputs a signal (V.sub.e) 107 to an
analog-to-digital converter 108 which outputs the digital
(response) signal, RIM.sub.ij 208, to a NUC system 110. The NUC
system then performs the process of noise reduction and
removing/replacing dead pixel data to help achieve target detection
and outputs digital signal CIM.sub.ij 112.
[0004] As shown in FIG. 1, the output image data signal from IR
sensor 102 is given by equation 104a where the sensor integration
time (IT), given by I.sub.p in the equation 104a, is a critical
parameter for producing a high magnitude image signal as input to
the amplifier 106. Switching to a longer sensor integration time
helps to produce a higher magnitude signal input to the amplifier
which aids noise reduction and increases sensor sensitivity leading
to early target detection and reliable target tracking and
recognition (identification). However, intelligent switching of the
integration time must occur since a longer integration time may
also lead to system (amplifier) saturation producing undesirable
non-linear effects.
[0005] Many current targeting systems employ IT switching
techniques that switch the integration time continuously on a
frame-by-frame basis to maintain input pixel intensity at a middle
intensity value to reduce starvation and saturation conditions for
the system. However, such frequent IT switching varies the sensor
sensitivity and requires more processing power. Additionally, such
frequent IT switching to a significant plurality of different
values increases calibration complexity for a targeting system when
measuring important parameters of a target object (e.g., measured
object irradiance needed for target discrimination and
classification) since a different calibration is required for each
operating IT. Additionally, raw pixel data output from the IR
sensor resulting in system saturation should not necessitate a
switch to a lower IT since a weak signature (e.g., low temperature)
target object may be obscured by nearby bright intensity (e.g.,
burning) counter-measurement (CM) objects or decoys that produce
the saturation condition. Under these conditions, a high or even
higher (increased) sensor sensitivity should be maintained and thus
the IT should not be switched to a lower value to eliminate the
saturation condition.
[0006] Therefore, due to the disadvantages of current IT switching
approaches, there is a need to provide a dynamic IT switching
system that maintains (produces) a high sensor sensitivity without
complicating important measurement calibrations and without
lowering the reliability of detecting a target object obscured by
uninterested objects that produce system saturation.
SUMMARY OF THE INVENTION
[0007] The method and system of the present invention overcome the
previously mentioned problems by providing a target detection and
tracking system capable of switching (changing) the integration
time (IT) for the system IR sensor within a discrete set of values
to maintain a high sensor sensitivity. The integration time is
dynamically changed to the same or a different sensor integration
time within the discrete set based on the image data output from
the sensor satisfying pre-determined system parameter thresholds.
Further features of the present invention include an IT-related
saturation prediction function allowing the system to avoid
unnecessary system saturation when determining whether an IT change
should be made. Additional features of the present invention
include a feedback function from the tracking portion of the system
that allows target objects with a low sensor signature to be
detected without being obscured by nearby uninterested objects that
produce system saturation.
BRIEF DESCRIPTION OF THE DRAWINGS
[0008] FIG. 1 is a block diagram of an exemplary target detection
system found in the prior art;
[0009] FIG. 2 is a functional block diagram of an exemplary
non-uniform correction system in accordance with an embodiment of
the present invention.
[0010] FIG. 3 is a block diagram of exemplary non-uniform
correction system with tracking feedback in accordance with an
embodiment of the present invention.
[0011] FIG. 4 shows a graph with exemplary sensor sensitivity
curves in relation to sensor background temperature ranges in
accordance with an embodiment of the present invention.
[0012] FIG. 5 shows a graph with exemplary predicted sensor
sensitivity curves as a function of temperature in accordance with
an embodiment of the present invention.
[0013] FIG. 6 is a flowchart of an exemplary integration time
switching algorithm in accordance with an embodiment of the present
invention.
[0014] FIG. 7 is an exemplary illustration of a burning
counter-object co-located with actual targets in accordance with an
embodiment of the present invention.
[0015] FIGS. 8-12 show graphs with exemplary performance sensor
sensitivity curves in accordance with an embodiment of the present
invention.
DETAILED DESCRIPTION
[0016] FIG. 2 shows a functional block diagram of an exemplary
non-uniform correction (NUC) system 200 in accordance with an
embodiment of the present invention. Advantageously, NUC system 200
may be incorporated into the target detection system 100 of FIG. 1,
replacing the prior art NUC system 110, to receive the digital
image signal (RIM.sub.ij) 208, carrying the raw image data
generated from sensor 102 (from the target signature) and output
from A/D converter 108, and generate output data for reliable
target detection by the system. The system 200 may include at least
three functional components including a permanent and blinking dead
pixel replacement (PBDPR) component 202, dynamic range management
(DRM) component 204, and non-uniform/fixed pattern noise (FPN)
Estimation and Removal (FPN E&R) component 206. Advantageously,
NUC system 200 may be a CWF (chopper-wheel-free) and MBPF
(Measurement-Based-Parametric-Fitt- ing) NUC system to better
discern background noise from raw image data output by sensor 102
and carried by digital image signal 208.
[0017] As described previously, the raw image data output by sensor
102 may include dead pixels (permanent and blinking) that are
either very bright or dark in intensity leading to non-linear
(e.g., saturation or starvation) conditions for the targeting
system resulting in low target detection reliability and lower
sensor sensitivity. To help reduce this problem, PBDPR functional
component 202 may receive input signal RIM.sub.ij 208 and perform
the function of replacing these dead pixels from input signal
RIM.sub.ij 208 (raw image data) using well-known global and local
thresholding techniques to generate output (response) signal
DPR.sub.ij.
[0018] As described in greater detail later, DRM functional
component 204 may receive input signal DPR.sub.ij 210 to initiate
the function of dynamically changing (switching) the integration
time (and electronic gain of amplifier 106) within a predetermined,
discrete set of values to maintain sufficient sensor sensitivity
(producing reliable target detection) within a linear dynamic
temperature range (for the sensor 102) divided into a multiple
number of predetermined operating temperature ranges that generate
raw image data, RIM.sub.ij 208. As a result of this process, DRM
component 204 generates output signal DRM.sub.ij 212 (equal to
input signal DPR.sub.ij 210).
[0019] FPN E&R functional component 206 may receive input
signal DRM.sub.ij 212, and perform the function of estimating and
removing the non-uniformity/fixed pattern noise (FPN) from input
signal DRM.sub.ij 212. The estimation and removal may
advantageously be based on lab measurements to determine the
particular parameter settings for greater noise estimation and
removal. Piece-wise (e.g., two pieces) or one-piece linear curve
fittings may be used for different predetermined temperature ranges
to estimate the FPN at different temperatures and ITs. As a result
of this process, DRM component 204 generates output signal
CIM.sub.ij 214.
[0020] As shown in FIG. 3, the NUC system 200 of FIG. 2 may
additionally include at least one advanced logic functional
component, a tracking/automatic target recognition (ATR) component
216. Tracking/ATR component 216 may receive input signal CIM.sub.ij
214 and perform the function of tracking and automatically
recognizing the detected target. As a result of this process,
Tracking/ATR component 216 may generate one or more output
(feedback) signals 218 including signals indicating whether
reliable tracking and/or ATR has been acquired or not.
Advantageously, digital signals of value 1 or 0 may be used to
indicate acquisition (e.g., Tracking=1, ATR=1) or no acquisition
(e.g., Tracking=0, ATR=0) of tracking and/or ATR. Output signals
218 are fed back into DRM component 204 to change NUC system 200
from a feedforward system to a feedback target detection system
that uses the feedback of tracking/ATR signals (carried by output
signal 218) to determine whether an IT switch should occur.
[0021] Advantageously, DRM component 204 may function as a
subsystem within NUC system 200 to adaptively (dynamically) adjust
the integration time (IT) of the sensor (FPA) to meet predetermined
requirements (thresholds) for sensor sensitivity and dynamic
temperature range (operating temperature) of the sensor.
Additionally, DRM component 204 may function to reduce the IT of
the sensor in response to pixel saturation. Advantageously, DRM
component 204 may be embodied as logic device using dynamic random
access memory (DRAM).
[0022] During operation, DRM component 204 may use an algorithm
(DRM algorithm) to switch the sensor IT for efficiently detecting a
plurality of targets even when the targets may be located among
high intensity counter-measurement objects (decoys) that create a
system saturation condition. Execution (operation) of the algorithm
may be predicated on the pre-selection of at least four integration
times corresponding to four temperature ranges of interest to
detect targets.
[0023] Advantageously, the integration times are pre-selected
(predetermined) to maintain high sensor sensitivity requirements
for each temperature range. For target temperatures greater (>)
than 70 degrees Celsius (.degree.C.) (saturated target intensities)
and background temperature range of -10.about.>70.degree. C., an
IT approximately equal to 2.5 milliseconds (ms) may be selected for
the sensor. A shorter IT is needed for this higher temperature
range for the target to avoid system saturation. Advantageously,
for lower target temperatures (<70.degree. C.) a plurality of
other integration times may be selected to maintain high sensor
sensitivity. For a background temperature range of
35.about.70.degree. C., an IT approximately equal to 5.2
milliseconds (ms) may be selected for the sensor. Alternatively,
for a background temperature range of 0.about.35.degree. C., an IT
approximately equal to 9 milliseconds (ms) may be selected for the
sensor. Further alternatively, for lower target temperatures
(<70.degree. C.) and a background temperature range of
-33.about.0.degree. C., an IT approximately equal to 16.6
milliseconds (ms) may be selected for the sensor. Advantageously,
the maximum amount of time for switching from the lower temperature
range IT to the highest temperature DR (e.g., 16.6 to 2.5 ms) by
the DRM algorithm may be a sampling duration of three frames.
[0024] Advantageously, sensor 102 may be embodied as a Cincinnati
Electronics Indium/Antimony (CE-InSb) focal plane array where the
sensor maintains high sensitivity for the different IT's
pre-selected and corresponding to the different background
temperature ranges. It is noted that the selection of a CE-InSb
sensor (FPA) and the associated IT's (based on the related
background temperature ranges) are solely exemplary and should not
be viewed as a limitation upon the present invention. As such,
alternative manufacturers and models of FPAs (sensors), along with
alternative integration times, may be selected and still be within
the scope of the present invention.
[0025] As shown in FIG. 4, for CE-InSb sensors, sensitivity at
particular background temperature ranges is related (inversely
proportional) to the NEDT (noise equivalent delta temperature) for
the sensor. These sensor sensitivity characteristics allow a low
temperature target to be detected with a lower NEDT (due to
increased sensitivity at lower NEDTs). The NEDT may be calculated
from the CE camera measurement data (e.g., F, F/4, and Tau) to
produce the sensitivity curves as shown in FIG. 4, where F is the
lens F-number, F/4 is the number for the CE camera, and Tau is the
optical transmittance. As shown in FIG. 4, at lower background
temperature ranges (e.g., -30.degree. C.), a higher NEDT is
produced which inversely decreases the sensor sensitivity
necessitating the use of a longer IT (e.g., 16.6 ms) to maintain
efficient target detection by the target detection system 200.
Alternatively, for higher background temperature ranges (e.g.,
60.about.70.degree. C.), a lower NEDT is produced which inversely
increases the sensor sensitivity and necessitates the use of a
shorter IT (e.g., 5.2 ms) to avoid system saturation from the
higher background temperature and maintain target detection by
system 200.
[0026] Advantageously, a plurality of predetermined system
requirements may be established for the sensor relating to
particular, calculated NEDT values and corresponding background
temperature ranges to maintain high sensor sensitivity for accurate
target detection. For example, it may be required that the maximum
background temperature (for the sensor) approximately equals
63.degree. C. From FIG. 4 (produced from lab measurements and
sensor data), it can be verified that for an IT approximately equal
to 5.2 ms, the background temperature range for the sensor is
0.degree. C..about.70.degree. C. ensuring that the maximum
background temperature requirement may be satisfied for all
IT.ltoreq.5.2 ms.
[0027] A second system requirement to be satisfied for greater
sensitivity may be to maintain NEDT.ltoreq.50 mk at 22.degree. C.
for all ITs.ltoreq.5.2 ms. From FIG. 4, for an IT approximately
equal to 5.2 ms, NEDT (from lab measurements) may be calculated to
be 35.5 mk (mili Kelvin) at 22.5.degree. C., where F=4 and
Tau=0.88. As shown in FIG. 5, sensor sensitivity at this background
temperature range (22.5.degree. C.) may be increased based on
changing the optical settings for the sensor. For example, changing
Tau to 0.522 and F to 2.47 reduces the NEDT to approximately 22.8
mk at 22.5.degree. C. These calculations verify that this second
system requirement (NEDT.ltoreq.50 mk at 22.degree. C.) may be
satisfied for all IT.gtoreq.5.2 ms.
[0028] A third system requirement may be to maintain a minimum
background temperature (for the sensor) of -33.degree. C. As shown
in FIGS. 4-5, for these sensors the NEDT increases significantly at
very low (background) temperatures (e.g., -30.degree. C.) producing
the inverse result of low FPA (sensor) sensitivity at low
temperatures. As shown in FIG. 5, NEDT is approximately equal to 46
mk at -28.degree. C. (where F=2.47 and Tau=0.522) which ensures
that this third system requirement (minimum background temperature
of -33.degree. C.) may be satisfied all ITs.gtoreq.16.6 ms.
[0029] Advantageously, the background temperature (relating to the
operation temperature of sensor 102) used in the DRM 204 algorithm
may relate to the specific image mean (averaged) value for images
received and output by sensor 102. The specific image mean may be
measured and defined as a mean image count (MIC) relating to the
pixel intensity for the received raw image data. During execution
of the algorithm, the MIC count (value may be given in COUNT units)
for images received by sensor 102 may be used to determine the
current background temperature range and the corresponding,
currently selected IT. For example, MIC.sub.--0.sub.--16 and
MIC.sub.--0.sub.--9 may be refer to the mean image count values for
background temperature ranges of 0.degree. C. (for each MIC values)
with corresponding ITs (approximately) equal to 16.6 and 9 ms,
respectively. The MIC count may be calculated from the center
64.times.64 elements of the FPA.
[0030] A key feature of the DRM algorithm for switching integration
times (example shown in the Appendix) performed by DRM component
204 is the ability to distinguish between saturated pixels caused
by high temperature targets (or objects). Advantageously, the DRM
algorithm may use a "SAT" variable to indicate when system
saturation has occurred or not (SAT=1 for saturation, or SAT=0 for
no saturation). For example, the algorithm may be designed to
produce SAT=1 if there are M pixels with Tao values of larger than
4050 (maximum Tao=4096), or otherwise SAT=0. M may be a selectable
parameter (e.g., 2.about.3% of the center 64.times.64 pixels), and
saturation pixels may be estimated by a fast histogram method
(example using MATLAB code shown in the Appendix) using a logic
device such as a field programmable gate array (FPGA). For a
tracking mode of the algorithm (determining whether tracking has
been acquired or not), the DRM component 202 may use the center
64.times.64 pixels to compute the histogram. Alternatively for the
ATR mode (determining whether ATR has been acquired or not), the
DRM component 202 may use the whole 256.times.256 pixels (elements)
from the image signal (RIM.sub.ij 208) generated from sensor 102.
Also, to reduce the computational intensity for the ATR mode,
under-sampling of the 256.times.256 image may be performed using
every 4.sup.th or 8.sup.th pixel.
[0031] Under some situations the saturated pixels may be from some
uninterested objects or clutters (decoys) and the signatures of a
target may still be very weak. Advantageously, under these
situations, the DRM component 204 may maintain a high sensitivity
mode (not switch to a shorter IT) using a feedback feature (from
output signals 218 from Tracking/ATR component 216) of the
algorithm.
[0032] Additionally, when SAT=0, performance of the DRM algorithm
(by DRM component 204) includes using a
saturation-integration-time-prediction (SRITP) function (example
using MATLAB shown in the Appendix) to determine whether a switch
to a longer IT may be performed to increase sensor sensitivity
without causing system saturation. The SRITP function may predict
the particular integration time (IT) that increases the system gain
to produce SAT=1 (if M pixels with Tao values larger than
Tao=4050). After the predicted IT is produced by the SRITP
function, DRM component 204 may compare the predicted IT with the
actual IT to be switched to and decide to do either of the
following: 1) if the predicted IT (causing system saturation)
equals the IT to be switched to, then do not switch to a longer IT
and maintain the current IT, or 2) if the predicted IT does not
equal (e.g., greater than) the IT to be switched to, then switch to
the longer IT to increase sensor sensitivity. The use of the SRITP
functions allows the DRM to work under a more steady state
operation without frequent switching of the sensor IT.
[0033] FIG. 6 is a flowchart illustration of the DRM algorithm
shown in Appendix A wherein signal DPR.sub.ij 210 (output signal
produced by PBDPR component 202) is received by DRM component 204
to initiate performance of the DRM algorithm which may produce
output (response) signal DRM.sub.ij 212. Advantageously, as shown
by expression 605, response signal DRM.sub.ij 212 is equal to input
signal DPR.sub.ij 210 as performance of the DRM algorithm does not
change the value of input signal 210, but uses this input to
control the sensitivity of sensor 102 by dynamically switching
sensor integration time (in response to predetermined criteria) for
future image time frames. Advantageously, B is an adjustable
buffering integer number (e.g., set at 100) that may be used to
prevent integration times from switching back and forth when the
background temperature may be close to the two IT-switching
temperatures (e.g., 0.degree. C. and 35.degree. C.). Also, when the
DRM algorithm is initially performed, the actual targets may be a
far distance away from the sensor 102 which makes it advantageous
to initiate the DRM algorithm, for a predetermined number of frames
(e.g., 5 frames), by selecting an IT corresponding to the middle
background temperature range (e.g., 9 ms for 0.about.35.degree. C.)
for the sensor (from sensor data generated from lab measurements).
This preselected IT/temperature range corresponds to mean image
counts MIC.sub.--0.sub.--9 (defined by low_bound) and
MIC.sub.--35.sub.--9 (defined by high_bound) which sets the initial
IT (for the DRM algorithm) to 9 ms for the corresponding background
temperature range of 0.about.35.degree. C.
[0034] As a preliminary step to performance of the DRM algorithm
(before actual image sensing by sensor 102) when saturation has not
occurred (e.g., SAT=0), the SRITP function is performed to predict
the sensor IT that will produce system saturation. At step 602,
image sensing begins and the mean image count (MIC) is measured for
the image signal (DPR.sub.ij) 210 input to DRM component 204 and
compared with the lower-end MIC value (low_bound) for IT=9 ms.
Advantageously, the mean image count value is measured for a
predetermined number of image frames (e.g., for n frames) to obtain
the best possible MIC value. If the measured MIC (MIC_cf) is less
than the lower-end MIC value (e.g., for IT=9 ms at 0.degree. C.),
then the algorithm proceeds to step 608 where the current IT value
is upwardly switched (incremented) to the next higher value for the
corresponding lower background temperature range (e.g., switched to
IT=16.6 ms for -33.about.0.degree. C.), and the algorithm proceeds
to step 614. Otherwise, if MIC_cf is not less than low_bound, then
the algorithm proceeds to step 604. At step 604, MIC_cf is compared
with high_bound (high end MIC value for IT=9 ms). If MIC_cf is
greater than high_bound, then the algorithm proceeds to step 610
where the current IT value is downwardly switched (decremented) to
the next lower value for the corresponding higher background
temperature range (e.g., switched to IT=5.2 ms for
35.about.70.degree. C.). Otherwise, if MIC_cf is not greater than
high_bound, then the algorithm proceeds to step 606.
[0035] Advantageously, steps 602, 604, 608, and 610 may constitute
the feed-forward portion (using basic logic functions) of the
target detection system 200 that uses the measurements of the input
image data to efficiently switch the current sensor IT to maintain
reliable target detection. The other steps (606, 612, 614, 616,
618, 620) of the DRM algorithm shown in FIG. 6 may constitute the
feedback portion (using advanced logic functions) of target
detection system 200 which use the results from the tracking/ATR
component 216 (e.g., output signals 218 indicating tracking=0 or 1,
ATR=0 or 1) to determine whether to increment, decrement, or
maintain the current sensor IT value to maintain high sensor
sensitivity leading to reliable target detection (of target
signature) within the dynamic temperature range of the sensor.
[0036] At step 606, if MIC_cf is not greater than high_bound,
SAT=1, and either Tracking=1 or ATR=1, then the algorithm proceeds
to step 612 where the current IT value is downwardly switched
(decremented) to the next lower value for the corresponding higher
background temperature range (e.g., switched to IT=5.2 ms for
35.about.70.degree. C.). For this portion of the algorithm, since
tracking and/or ATR has been acquired, the algorithm switches to a
shorter IT to eliminate the system saturation condition.
[0037] Alternatively at step 606, if SAT=1, but tracking and ATR=0,
then most likely the system saturation (saturated pixels) is being
caused by uninterested objects (e.g., burning counter-measurement
objects, burning decoys, clutter) and therefore the DRM algorithm
will maintain the current, longer, and higher-sensitivity IT to try
and detect the weak target among the bright intensity background
clutter.
[0038] At step 614, if system saturation exists (SAT=1) and
tracking or ATR has been acquired (tracking=1 or ATR=1), then the
algorithm proceeds to step 620 where the current IT value is
downwardly switched (decremented) to the next lower value for the
corresponding higher background temperature range. For this portion
of the algorithm, since the original sensor IT was previously
incremented at step 608, the original IT is effectively
re-established (maintained) by now reversing the previous increase
(by decrementing in step 620). Since tracking=1 and/or ATR=1, the
algorithm wants to maintain the current tracking and/or ATR
acquisition of the target despite the current system saturation
condition. Otherwise, if system saturation does not exist (SAT=0),
then the algorithm proceeds to step 616.
[0039] Alternatively at step 614, if SAT=1, but tracking and ATR=0,
then most likely the system saturation (saturated pixels) is being
caused by uninterested objects (e.g., burning counter-measurement
objects, burning decoys, or clutters) and therefore the DRM
algorithm will switch to the longer, higher-sensitivity IT to try
and detect the weak target among the bright intensity background
clutter.
[0040] FIG. 7 is an exemplary illustration, at various ranges to
the target, of the alternative scenarios described in relation to
steps 606, 614 where two burning decoys (barrels) 702, 704 may
attempt to obscure detection of the actual target (tank) 706
co-located (in between) with the decoys.
[0041] At step 616, if SAT=0 and the IT to be switched to is
greater than the predicted IT to produce system saturation (from
SRITP function), then the algorithm proceeds to step 618 where the
current IT value is downwardly switched (decremented) to the next
lower value for the corresponding higher background temperature
range. For this portion of the algorithm, it is determined that
switching to the next higher IT will result in system saturation.
Therefore, the algorithm decides to reverse the previous IT
increment (at step 608) and maintain the original IT value (e.g.,
IT=9 ms) to prevent system saturation with a switch. This decision
to stay at the original IT value helps to keep the system 200 in a
steady state operation without constant fluctuation of the IT
value. Advantageously, the steps of the DRM algorithm are repeated
indefinitely until an actual target hit occurs (or alternatively
the target detection system 200 is rendered inoperative).
[0042] FIGS. 8-12 illustrate various performance results
(sensitivity curves and recognition/detection vs. range curves)
using the DRM algorithm described herein in accordance with
embodiments of the present invention. N50 and N90 represent 50%
probability target detection and 90% probability target detection,
respectively. CE N50 represents the 50% probability target
detection as given by CE standards for the sensor (camera).
Advantageously, during operation for a low background temperature
example (e.g., -33.about.0.degree. C.) as shown in FIGS. 8-10, the
DRM algorithm may switch from an (initially set) IT=9 ms to IT=16.6
ms after the fifth frame (e.g., SAT=0 and switched to
IT.ltoreq.IT_predict, and tracking=1 and/or ATR=1, from step 616)
and stay there until an actual target hit occurs. Alternatively,
during operation for a mid-background temperature example (e.g.,
0.about.35.degree. C.) as shown in FIG. 12, the DRM algorithm may
continually stay at an (initially set) IT=9 ms after the fifth
frame (e.g., SAT=0 and switched to IT>IT_predict, and tracking=1
and/or ATR=1, from step 616) and stay there until an actual target
hit occurs.
[0043] Also, alternatively, during operation for a target
pixel-saturation example (e.g., background temperature range of
-33.about.0.degree. C. and target temperature=65.degree. C.) as
shown in FIG. 11, the DRM algorithm may switch from an (initially
set) IT=9 ms to IT=16.6 ms after the fifth frame (e.g., SAT=0, and
tracking=1 and/or ATR=1, from step 616), then switch back to IT=9
ms a few frames later (e.g., SAT=1, and tracking=1 and/or ATR=1,
from step 606), and then switch to IT=5.2 ms (e.g., SAT=1 and
tracking=1 and/or ATR=1, from step 606) and stay there until an
actual target hit occurs.
[0044] Further alternatively, during operation for a burning
counter-measurement example (for various ranges--R) as shown in
FIG. 7 (e.g., burning decoys at 100.degree. C. causing SAT=1
although background temperature range of 0.about.35.degree. C. and
target temperature=25.degree. C.), the DRM algorithm may select an
(initially set) IT=9 ms (with high sensitivity) after the fifth
frame (e.g., SAT=1, but tracking=0 and/or ATR=0, from step 606) and
then maintain IT=9 ms (e.g., SAT=1, and tracking=0 and/or ATR=0,
from step 606) and stay there until an actual target hit
occurs.
[0045] A NUC system using the DRM algorithm described herein, in
accordance with embodiments of the present invention, provides a
number of advantages. These advantages include the division of the
required dynamic temperature range (e.g., -33.about.63.degree. C.)
of the system sensor into three sub-dynamic ranges corresponding to
three pre-selected (predetermined) sensor IT's (e.g. 5.2, 9, 16.6
ms) which helps the NUC system maintain requirements for operation
temperature dynamic range and sensor (system) sensitivity.
Additionally, a NUC system using the DRM algorithm may also select
an IT (e.g., 2.5 ms) to avoid pixel saturation for a very hot
target (or a close target). And with additional information from
conditions such as saturation, tracking, and ATR, the NUC system
can adaptively (dynamically) select the appropriate IT's to
maintain target intensity within a given linear dynamic range (even
for a hot target).
[0046] Although the invention is primarily described herein using
particular embodiments, it will be appreciated by those skilled in
the art that modifications and changes may be made without
departing from the spirit and scope of the present invention. As
such, the method disclosed herein is not limited to what has been
particularly shown and described herein, but rather the scope of
the present invention is defined only by the appended claims.
* * * * *